.published-loading { display: none; }

Mistral 7B Instruct v0.2

Mistral AI · 7B parameters · browser LLM configuration and published measurements.

Runtime and model configuration

Configured weight formats: 4-bit weights · FP16 compute. Inference uses WebLLM and WebGPU in a dedicated browser worker.

Publisher model card · Run benchmark

Published benchmark results

q4f16_1 · WebLLM 0.2.82

Median generation speed: 51.9 tok/s
Median first-output latency: 223.0 ms
Median response time: 1.44 s
Sampled peak Chrome footprint: 7.08 GB
Completed measured requests: 1
Browser and device: Chrome 155.0.0.0 · M5 Pro · 48 GB · dedicated Chrome

Download JSON report

Warmup is excluded from performance medians. Tokenizers and output lengths differ across models. These measurements do not evaluate task quality. Measurement definitions